Determinants of life satisfaction among Ghanaians aged 15 to 49 years: A further analysis of the 2017/2018 Multiple Cluster Indicator Survey.

The inclusion of life satisfaction in government policies as a tracker of the social and economic progress of citizens has been recommended. This has encouraged the scientific investigation of life satisfaction levels of people in tandem with factors responsible for these levels. Only a few studies have attempted to do this in Ghana with mixed findings. This study, therefore, extends previous literature by examining the determinants of life satisfaction among Ghanaians in two ways: a full sample and a gender-stratified sample. We analysed cross-sectional data from the 2017/2018 Ghana Multiple Indicator Cluster Survey Six (MICS 6). A sample of 20,059 women and men of ages ranging from 15 to 49 years participated in this study. The Cantril's Self-Anchoring Ladder Life Satisfaction scale was used to capture the life satisfaction of participants alongside relevant sociodemographic questions. About 35% of participants reported they were satisfied in life with males reporting more suffering levels [39.59%; 95% CI:36.38, 42.88] and females more thriving levels [36.41%; 95% CI:35.01, 37.84]. In the full sample multivariable model, gender, age, parity, education, marital status, wealth index, and region of residence were significantly associated with life satisfaction. Gender variations were also found across these associations. These findings collectively provide useful information for policymakers and practitioners to optimize interventions for the Ghanaian population aimed at improving life satisfaction. Evidence from this study also calls on the government of Ghana to begin tracking the life satisfaction of her citizens.

Introduction

Life satisfaction refers to the subjective evaluation of life as a whole [1]. It implies that life satisfaction is a subjective cognitive appraisal of an individual’s condition regarding several life domains [2]. Researchers have expressed differing perspectives on the domains that comprise life satisfaction. Some scholars suggested that life satisfaction is dependent on a combination of both material and non-material factors such as one’s health status, social relationship, family, and self-worth [3,4]. Others conceptualize life satisfaction based solely on material factors such as work, income, and place of residence [5,6].

Despite the divergent views about its domain, there seems to be a consensus in literature on the associated factors of life satisfaction from different parts of the world. In developed countries, factors such as household and personal income, health, age, gender, marital status, and education [5,7-12] have been identified as significant associates of life satisfaction. In Diego-Rosell, Tortora, and Bird’s [9] study, for instance, it was found that families with a high household income had better life satisfaction. Luhmann et al. [11] also found from analysing longitudinal data from three nationally representative panel studies that life satisfaction levels increased with marriage and childbirth but reduced with marital separation, job loss, starting a new job, and relocating.

Similar findings on associated factors have been reported in studies from Sub-Saharan Africa, particularly South Africa, Malawi, Ethiopia, Nigeria, and Ghana [13-17]. In the Ghanaian context, the few existing evidence shows that being involved in religion, experiences of migration, having a high income, higher education, social capital, tobacco use, being in the lower class, residing in the southern parts, job security and being married were related to life satisfaction [13,18-26].

While these few Ghanaian studies provide preliminary evidence, they also suggest the need for more investigation into the factors associated with life satisfaction. Moreover, the majority of these studies have focused on only one aspect or domain of life satisfaction. For instance, Pokimica et al. [21], as well as Addai and Pokimica [18], operationalized life satisfaction in terms of the living standards of individuals rather than how individuals feel about their overall life. So far, only Addai et al. [13] and Calys-Tagoe et al. [20], the most recent studies in Ghana, have measured life satisfaction as the extent to which individuals feel about their overall life. Even though both studies relied on a nationally representative sample of men and women, there are some methodological limitations worth mentioning. The age of Addai et al’s [18] dataset is now quite old, using data that was collected within 2005–2008. Also, Calys-Tagoe et al’ study [20] only focused on older adults (50 years and above).

Considering these limitations and the pressing need to produce more recent evidence in particular because of persisting contextual problems ranging from limited access to drinking water [27], unemployment particularly among the youth [28], limited access to health care, poverty [29], high prevalence of chronic diseases [30-34] and poor quality education [35,36] that may threaten one’s life satisfaction, our study used the 2017/2018 Multiple Cluster Indicator Survey to examine the factors associated with life satisfaction in Ghana. Our study goes a step further by examining closely these factors from a gendered perspective; an examination that is virtually non-existent in Ghana. Due to the variations in social norms and biological characteristics, a gendered perspective has been recommended to provide more nuanced information into the associated factors of both men and women’s life satisfaction [37-39]. The outcome of such attempt has been highlighted by Joshanloo [37] who concluded that socio-political, employment-related, and education-related variables were more important to men’s life satisfaction whereas women’s life satisfaction was associated with variables related to marital status and interpersonal relationships. It is, therefore, reasonable to expect that such unique differences may exist in our study. The findings of this study will be useful for policymakers, researchers, and practitioners in designing gender-specific interventions and services to improve the life satisfaction of men and women in Ghana.

Methods

Study design and data source

This study used existing data from the 2017–2018 Ghana Multiple Indicator Cluster Survey Six (MICS 6). The Ghana MICS is a cross-sectional survey conducted by the Ghana Statistical Service (GSS) in collaboration with the Ghana Health Service (GHS), Ministry of Health (MOH), and the Ministry of Education with funding and technical support being provided by UNICEF and other international donors [40]. In the 1990s, UNICEF launched the Global MICS Programme as an international multi-purpose household survey initiative to assist countries in gathering internationally comparable data on a wide range of initiatives on the situation of children and women. MICS analyses key indicators that assist countries to generate data for use in national development plans, policies, and programmes, as well as to measure progress towards SDGs and other agreements signed internationally [40].

A multi-stage, stratified cluster sampling approach was used to nationally survey children and women in urban and rural areas, located in the previously 10 administrative regions in Ghana: Western, Central, Greater Accra, Volta, Eastern, Ashanti, Brong Ahafo, Northern, Upper East and Upper West [40]. The sampling frame for data collection was based on the 2010 Population and Housing Census (2010 PHC) of Ghana. At the first stage guided by the definition of the 2010 PHC enumeration, the enumeration areas (EAs) were identified within the selected primary sampling units (PSUs). In each EA sample, the cataloguing of households was carried out and a sample of households was selected in the second stage using systematic random sampling. In each sampled household, all persons who met the inclusion criteria (e.g., age 15–49 years) were eligible to participate in the survey. The smallest regions were allocated a minimum of 60 sample clusters (primary sampling units), and the Greater Accra Region was allocated a maximum of 86 sample clusters. The sample clusters were distributed between the urban and rural strata within each area, proportionate to the size of the corresponding populations within the frame. Clusters (primary sampling units) were assigned to the urban and rural strata in each area in proportion to the number of households in the census frame for each stratum within that region [40]. The final samples were 660 clusters and 13202 households across all sampling strata.

Ethics

UNICEF through the Ghana Statistical Service obtained ethical clearance. Verbal consent and assent were obtained from participants age 18+ and 15-17years, respectively. All participants were informed about the voluntary nature of participation including confidentiality and anonymity.

Study sample

Our full sample size totalled 20,059 women and men of ages ranging from 15 to 49years. Women and men numbered 14,587 and 5,472, respectively.

Measures

On behalf of the GSS and UNICEF, trained enumeration officials collected the data. Six questionnaires were included in the field data collection instrument: 1) Household questionnaire, 2) Water Quality Testing Questionnaire, 3) Questionnaire for Individual Women, 4) Questionnaire for Individual Men, 5) Questionnaire for Children Under Five, and 6) Questionnaire for Children Age 5–17. We used data collected with Individual Women and Individual Men questionnaires that were administered to randomly selected women and men living in the surveyed household.

Outcome variable

The outcome variable, life satisfaction, was measured using the Cantril’s Self-Anchoring Ladder of Life Satisfaction scale in the MICS. An image of a ladder with steps numbered from ‘0’ at the bottom to ‘10’ at the top was shown to participants and they were asked to indicate at which step of the ladder they believed they stood about their level of life satisfaction at the time of the survey. Following the recommendation of Cantril [41] and Gallup [42], the responses were categorized as suffering (0–4) as ‘0’, struggling (5–6) as ‘1’, and thriving (7–10) as ‘2’. This recategorized variable was kept solely for the purposes of description; the original variable (i.e., the 0 to 10 ordinal measurement) was used in the main study analyses.

Predictor variables

The following variables were selected as predictor variables: gender, age, marital status, education, happiness level, health insurance, rural-urban, household wealth, and region of residence. Our selection was based on reports from prior research and variable availability [13,20]. We maintained the original categorization of each of these variables from the dataset. See Table for the categorization. Detailed explanations for these variables are provided elsewhere [40]. Briefly, most of the selected variables were measured in a simple manner (e.g., “Are you covered by any health insurance?” with responses “Yes” or “No”) while others were aggregated from responses to several questions like the computation of household wealth of participants using household characteristics, possessions and assets (e.g., internet access, number of rooms for sleeping, source of drinking water, ownership of television, radio, vehicles and access to electricity, among others). Household wealth was categorized into poorest (0), second quintile (1), middle (2), fourth quintile (3), and richest (4).

Data preparation and analysis

Data analysis began by recoding the selected variables after appending both male and female datasets in Stata version 14. Spearman’s rho was used as a preliminary test to assess the intercorrelations between all study variables (see S1 Table). Next, we weighted the data to perform univariable analysis computing for frequency distributions of study variables. We accounted for the complex sampling design embedded in the dataset to account for possible analytical errors and make proper inferences [43]. This was achieved by correcting for clusters, stratification, and sample weights using the complex survey mode command ’svyset’. After this correction, we performed bivariable (see S2 Table) and multivariable analyses, regressing life satisfaction (as it was originally measured with the Cantril’s ladder) onto the predictor variables on the full and gender-stratified samples using ordered probit regression (‘oprobit’) command. The ordered probit model is typically used to examine the variation in the data points of an ordinal categorical dependent variable [44]. Though argued to produce parameter estimates difficult to interpret, oprobit was fitted mainly for its ability to preserve the ordering of the response options in the outcome variable as a function of the predictor variables [45]. Next, we ran the margins command to produce predicted probabilities (see Table 3) only for the gender-stratified models and to ease interpretation of the estimated coefficient from the oprobit output [44]. Additionally, margins plots were generated for the highest level of life satisfaction to further support the interpretation of the predicted probabilities. We decided to make the plots more compact by interacting gender with selected predictors including age groups, education level, household wealth index and marital status using the full sample model (see S1 Fig). We report only adjusted models, pegging statistical significance at p≤0.05.

Table 3

Predicted probabilities for male and female sample at 3 points (lowest, middle and highest) of life satisfaction ladder.

	Male (Margins [Std. E])			Female (Margins [Std. E])
Predictors	Lowest (0)	Middle (5)	Highest (10)	Lowest (0)	Middle (5)	Highest (10)
Age
15–19 years	.017*** (.003)	.195*** (.010)	.071*** (.010)	.023*** (.003)	.213*** (.006)	.121*** (.009)
20–24 years	.025*** (.003)	.195*** (.010)	.051*** (.006)	.029*** (.003)	.217*** (.006)	.102*** (.007)
25–29 years	.027*** (.004)	.194*** (.010)	.048*** (.007)	.029*** (.003)	.217*** (.006)	.101*** (.006)
30–34 years	.020*** (.003)	.195*** (.010)	.060*** (.008)	.023*** (.003)	.213*** (.007)	.121*** (.013)
35–39 years	.020*** (.005)	.195*** (.010)	.061*** (.009)	.025*** (.003)	.214*** (.006)	.115*** (.007)
40–44 years	.017*** (.004)	.195*** (.010)	.068*** (.009)	.030*** (.005)	.217*** (.006)	.102*** (.009)
45–49 years	.015*** (.004)	.194*** (.010)	.074*** (.010)	.020*** (.003)	.210*** (.006)	.131*** (.009)
Education
Pre-primary or none	.022*** (.006)	.196*** (.010)	.056*** (.011)	.028*** (.003)	.217*** (.007)	.105*** (.010)
Primary	.021*** (.003)	.196*** (.010)	.057*** (.007)	.025*** (.003)	.215*** (.006)	.111*** (.007)
Junior Secondary	.019*** (.003)	.196*** (.010)	.064*** (.006)	.026*** (.003)	.216*** (.006)	.108*** (.005)
Senior Secondary	.023*** (.004)	.195*** (.010)	.053*** (.005)	.025*** (.003)	.215*** (.006)	.113*** (.007)
Higher	.011*** (.002)	.191*** (.010)	.092*** (.011)	.014*** (.002)	.198*** (.007)	.165*** (.011)
Marital Status
Currently married/in union	.019*** (.003)	.196*** (.010)	.063*** (.007)	.021*** (.002)	.210*** (.006)	.130*** (.006)
Formerly married/in union	.047** (.019)	.183*** (.019)	.027** (.011)	.036*** (.004)	.220*** (.006)	.086*** (.007)
Never married/in union	.019*** (.003)	.196*** (.010)	.064*** (.007)	.031*** (.003)	.219*** (.006)	.094*** (.007)
Parity
No child	.020*** (.003)	.195*** (.010)	.063*** (.007)	.022*** (.002)	.210*** (.006)	.130*** (.008)
One child	.023*** (.005)	.195*** (.010)	.056*** (.010)	.025*** (.003)	.215*** (.006)	.113*** (.007)
Two children	.022*** (.004)	.195*** (.010)	.058*** (.010)	.030*** (.004)	.217*** (.006)	.098*** (.007)
Three children	.019*** (.004)	.195*** (.010)	.065*** (.011)	.029*** (.003)	.217*** (.006)	.102*** (.008)
Four children	.018*** (.004)	.195*** (.010)	.065*** (.009)	.027*** (.003)	.216*** (.006)	.107*** (.007)
Insurance coverage
Without insurance	.022*** (.003)	.196*** (.010)	.054*** (.005)	.026*** (.002)	.215*** (.006)	.111*** (.006)
With insurance	.015*** (.002)	.195*** (.010)	.073*** (.006)	.025*** (.002)	.214*** (.006)	.114*** (.005)
Household wealth index
Poorest	.019*** (.003)	.198*** (.010)	.058*** (.007)	.032*** (.004)	.220*** (.006)	.092*** (.008)
Second	.026*** (.005)	.196*** (.011)	.045*** (.006)	.033*** (.004)	.220*** (.006)	.089*** (.007)
Middle	.026*** (.004)	.196*** (.010)	.045*** (.006)	.027*** (.003)	.218*** (.006)	.104*** (.008)
Fourth	.019*** (.003)	.198*** (.010)	.058*** (.007)	.023*** (.003)	.214*** (.006)	.119*** (.007)
Richest	.010*** (.002)	.190*** (.010)	.098*** (.010)	.016*** (.002)	.204*** (.006)	.150*** (.008)
Rural-Urban
Rural	.019*** (.002)	.194*** (.010)	.067*** (.007)	.026*** (.003)	.215*** (.006)	.111*** (.007)
Urban	.022*** (.004)	.194*** (.010)	.059*** (.006)	.025*** (.003)	.214*** (.006)	.114*** (.006)
Region of residence
Western	.021*** (.005)	.198*** (.010)	.054*** (.011)	.037*** (.005)	.221*** (.006)	.081*** (.008)
Central	.041*** (.007)	.187*** (.011)	.028*** (.006)	.031*** (.004)	.219*** (.006)	.094*** (.008)
Greater Accra	.011*** (.002)	.193*** (.010)	.090*** (.012)	.024*** (.003)	.216*** (.006)	.112*** (.008)
Volta	.010*** (.002)	.192*** (.010)	.093*** (.012)	.018*** (.004)	.207*** (.007)	.141*** (.016)
Eastern	.025*** (.004)	.197*** (.010)	.046*** (.006)	.024*** (.003)	.215*** (.007)	.113*** (.010)
Ashanti	.022*** (.005)	.198*** (.011)	.050*** (.008)	.034*** (.004)	.220*** (.006)	.088*** (.007)
Brong Ahafo	.017*** (.004)	.198*** (.010)	.062*** (.010)	.020*** (.003)	.211*** (.007)	.130*** (.010)
Northern	.011*** (.003)	.194*** (.011)	.087*** (.014)	.016*** (.003)	.203*** (.007)	.152*** (.012)
Upper East	.018*** (.005)	.198*** (.010)	.060*** (.013)	.007*** (.001)	.173*** (.008)	.231*** (.018)
Upper West	.009** (.004)	.190*** (.012)	.099*** (.024)	.014*** (.002)	.200*** (.007)	.162*** (.012)

Note.

* p < 0.05

** p < 0.01

*** p < 0.001

Robust standard errors are in parentheses.

Results

The descriptive results showed that about 35% [95% CI:33.49, 35.95] of all respondents are in the thriving category of the Cantril’s Self-Anchoring Ladder of Life Satisfaction scale. Nonetheless, there are some gender differences. As shown in Fig 1, about 40% [95% CI:36.38, 42.88] of males and 31% [95% CI:29.3, 32.25] of females reported that they were suffering while 30% [95% CI:27.85, 32.5] of males and 36% [95% CI:35.01, 37.84] of females reported that they were thriving. This interestingly suggests that more males are suffering than females and on the contrary, more females than males are thriving. The majority of respondents have obtained a middle school level education (8044/19670). About 51% have health insurance (10136/19670). Detailed information on the study variables is presented in Table 1.

Fig 1

Gender distribution of life satisfaction in Ghana.

Table 1

Summary statistics of study variables.

	Full sample	Male sample	Female sample
Variables	n (%)	n (%)	n (%)
Life satisfaction
[0] 0–4: Suffering	6520 (33.14)	2106 (39.59)	4414 (30.76)
[1] 5–6: Struggling	6323 (32.15)	1611 (30.29)	4712 (32.83)
[2] 7–10: Thriving	6828 (34.71)	1603 (30.13)	5225 (36.41)
Gender
[0] Male	5323 (27.0)
[1] Female	14374 (73.0)
Age
[0] 15–19 years	4413 (22.4)	1487 (27.93)	2926 (20.36)
[1] 20–24 years	3106 (15.8)	911 (17.2)	2195 (15.27)
[2] 25–29 years	2724 (13.8)	568 (10.68)	2156 (15.00)
[3] 30–34 years	2795 (14.2)	647 (12.16)	2148 (14.94)
[4] 35–39 years	2550 (12.9)	617 (11.59)	1933 (13.45)
[5] 40–44 years	2256 (11.5)	557 (10.46)	1699 (11.82)
[6] 45–49 years	1852 (9.4)	535 (10.05)	1317 (9.16)
Education
[0] Pre-primary or none	3228 (16.4)	525 (9.86)	2703 (18.81)
[1] Primary	3141 (15.9)	633 (11.89)	2508 (17.45)
[2] Junior Secondary	8044 (40.8)	2280 (42.84)	5764 (40.10)
[3] Senior Secondary	3948 (20.0)	1382 (25.95)	2566 (17.86)
[4] Higher	1335 (6.8)	504 (9.46)	831 (5.78)
Marital status
[1] Never married/in union	7526 (38.2)	2724 (51.17)	4802 (33.41)
(2] Currently married/in union	10606 (53.8)	2402 (45.12)	8204 (57.08)
[0] Formerly married/in union	1565 (7.9)	198 (3.72)	1367 (9.51)
Parity
[0] No child	7225 (36.68)	2818 (52.94)	4407 (30.66)
[1] One child	2248 (11.41)	419 (7.88)	1829 (12.72)
[2] Two children	2203 (11.18)	443 (8.33)	1760 (12.24)
[3] Three children	2094 (10.63)	419 (7.87)	1675 (11.65)
[4] Four + children	5927 (30.09)	1223 (22.98)	4704 (32.72)
Health Insurance coverage
[0] Without insurance	9561 (48.5)	3182 (59.78)	6379 (44.38)
[1] With insurance	10136 (51.5)	2141 (40.22)	7995 (55.62)
Household wealth index
[0] Poorest	3370 (17.1)	969 (18.21)	2401 (16.70)
[1] Second	3534 (17.9)	870 (1.34)	2664 (18.54)
[2] Middle	4020 (20.4)	1106 (20.78)	2914 (20.27)
[3] Fourth	4243 (21.5)	1202 (22.59)	3041 (21.16)
[4] Richest	4529 (23.0)	1176 (22.08)	3353 (23.33)
Rural-Urban residence
[0] Rural	9896 (50.2)	2811 (52.81)	7085 (49.29)
[1] Urban	9801 (49.8)	2512 (47.19)	7289 (50.71)
Region of residence
[0] Western	1940 (9.8)	520 (9.77)	1420 (9.88)
[1] Central	1867 (9.5)	459 (8.63)	1407 (9.79)
[2] Greater Accra	2531 (12.8)	642 (12.06)	1889 (13.14)
[3] Volta	1531 (7.8)	426 (8.01)	1105 (7.68)
[4] Eastern	2401 (12.2)	680 (12.77)	1721 (11.97)
[5] Ashanti	4744 (24.1)	1305 (24.52)	3439 (23.93)
[6] Brong Ahafo	1787 (9.1)	472 (8.87)	1315 (9.15)
[7] Northern	1839 (9.3)	517 (9.72)	1322 (9.20)
[8] Upper East	590 (3.0)	164 (3.08)	426 (2.96)
[9] Upper West	467 (2.4)	137 (2.57)	330 (2.30)

Associated factors of life satisfaction in a gender-stratified and full sample multivariable models

In the full sample multivariable model, gender, age, education, marital status, parity, wealth index, and region of residence were significantly associated with life satisfaction (see Table 2). Differences existed in the determinants of life satisfaction in the gender-stratified sample models. The association between age and life satisfaction was relatively the same for both males and females; the results revealed that males age group of 25–29 and females within the age groups of 20–24, 25–29, and 40–44 years had a reduced probability of reporting higher life satisfaction compared with those within the age group of 45–49 years. Education was also associated with life satisfaction. The findings revealed that having a higher education status increases males’ and females’ probability of reporting higher life satisfaction levels. The association between marital status, parity, household wealth index, region of residence, and life satisfaction was different for females and males. For instance, the probability of reporting higher life satisfaction was higher for women who were currently married compared with previously married women. For men, both those who were currently married/in union and never married/in union were associated with an increase in the probability of reporting higher life satisfaction levels. Regarding parity, women who had one child and more had a reduced probability of being satisfied with life compared with women with no children. However, parity was not statistically significantly related to life satisfaction among males. Higher life satisfaction levels were also reported by males who were covered by health insurance than males without insurance, although this association was not significant for females. Males belonging to the richest wealth quintile while females belonging to both the fourth and richest wealth quintile had an increased probability of reporting higher life satisfaction levels compared with those in the poorest wealth quintile.

Table 2

Multivariate oprobit model regressing life satisfaction on predictor variables in the full sample and gender stratified sample.

	Full Sample	Male sample	Female sample
Predictors	Coef. [95% CI]	Coef. [95% CI]	Coef. [95% CI]
Gender
Male	[ref]	NA	NA
Female	0.23*** [0.16, 0.29]
Age
15–19 years	-0.03 [-0.14, 0.081]	-0.02 [-0.25, 0.21]	-0.05 [-0.18, 0.08]
20–24 years	-0.16** [-0.26, -0.06]	-0.19 [-0.39, 0.004]	-0.16** [-0.27, -0.04]
25–29 years	-0.18*** [-0.27, -0.09]	-0.23* [-0.41, -0.04]	-0.16** [-0.26, -0.05]
30–34 years	-0.07 [-0.17, 0.03]	-0.11 [-0.26, 0.04]	-0.05 [-0.17, 0.07]
35–39 years	-0.10* [-0.18, -0.017]	-0.11 [-0.30, 0.08]	-0.08 [-0.17, 0.01]
40–44 years	-0.14** [-0.24, -0.035]	-0.05 [-0.21, 0.11]	-0.15* [-0.27, -0.04]
45–49 years	[ref]	[ref]	[ref]
Education
Pre-primary or none	[ref]	[ref]	[ref]
Primary	0.03 [-0.06, 0.13]	0.01 [-0.22, 0.24]	0.04 [-0.10, 0.17]
Junior Secondary	0.03 [-0.05, 0.11]	0.07 [-0.14, 0.27]	0.02 [-0.09, 0.13]
Senior Secondary	0.02 [-0.07, 0.11]	-0.03 [-0.24, 0.18]	0.05 [-0.07, 0.16]
Higher	0.28*** [0.17, 0.39]	0.27* [0.038, 0.51]	0.29*** [0.14, 0.44]
Marital Status
Currently married/in union	0.26*** [0.18, 0.34]	0.41* [0.062, 0.76]	0.24*** [0.17, 0.32]
Formerly married/in union	[ref]	[ref]	[ref]
Never married/in union	0.12* [0.001, 0.23]	0.42* [0.04, 0.81]	0.05 [-0.06, 0.17]
Parity
No child	[ref]	[ref]	[ref]
One child	-0.09* [-0.17, -0.004]	-0.06 [-0.27, 0.15]	-0.09* [-0.18, -0.001]
Two children	-0.14** [-0.23, -0.04]	-0.04 [-0.25, 0.18]	-0.17** [-0.28, -0.06]
Three children	-0.10* [-0.20, -0.01]	0.02 [-0.20, 0.24]	-0.15* [-0.26, -0.03]
Four + children	-0.08 [-0.17, 0.02]	0.02 [-0.17, 0.22]	-0.12* [-0.23, -0.01]
Insurance coverage
Without insurance	[ref]	[ref]	[ref]
With insurance	0.06** [0.02, 0.11]	0.16*** [0.08, 0.25]	0.02 [-0.03, 0.07]
Household wealth
Poorest	[ref]	[ref]	[ref]
Second	-0.05 [-0.14, 0.04]	-0.13 [-0.27, 0.01]	-0.02 [-0.13, 0.09]
Middle	0.02 [-0.07, 0.11]	-0.13 [-0.27, 0.02]	0.07 [-0.04, 0.18]
Fourth	0.12* [0.03, 0.20]	-0.004 [-0.15, 0.15]	0.15** [0.04, 0.26]
Richest	0.29*** [0.20, 0.39]	0.28** [0.11, 0.45]	0.30*** [0.19, 0.41]
Rural-Urban
Rural	[ref]	[ref]	[ref]
Urban	-0.003 [-0.07, 0.06]	-0.07 [-0.22, 0.08]	0.02 [-0.06, 0.10]
Region of residence	0	0	0
Western	[ref]	[ref]	[ref]
Central	-0.01 [-0.13, 0.11]	-0.32* [-0.58, -0.05]	0.08 [-0.05, 0.21]
Greater Accra	0.21*** [0.11, 0.32]	0.27* [0.04, 0.51]	0.19** [0.06, 0.31]
Volta	0.31*** [0.16, 0.46]	0.29* [0.047, 0.54]	0.33*** [0.15, 0.50]
Eastern	0.12* [0.01, 0.23]	-0.09 [-0.31, 0.14]	0.19** [0.06, 0.33]
Ashanti	0.03 [-0.09, 0.14]	-0.04 [-0.29, 0.21]	0.05 [-0.08, 0.17]
Brong Ahafo	0.22*** [0.11,0.34]	0.08 [-0.17, 0.32]	0.28*** [0.15, 0.41]
Northern	0.35*** [0.22, 0.49]	0.25 [-0.02, 0.53]	0.38*** [0.23, 0.52]
Upper East	0.51*** [0.37, 0.65]	0.06 [-0.23, 0.35]	0.67*** [0.52, 0.83]
Upper West	0.40*** [0.26, 0.55]	0.33 [-0.02, 0.68]	0.42*** [0.28, 0.56]
F statistics	F (32, 609) = 17.91, p<0.001	F (31, 610) = 8.36, p<0.001	F (31, 610) = 11.79, p<0.001

Note.

* p < 0.05

** p < 0.01

*** p < 0.001

95% CI: Confidence interval; Coef.: Robust regression coefficient.

There were also variations in the association between region of residence and life satisfaction among males and females. First, in both the male and the female samples, residing in Volta and Greater Accra regions were associated with a higher probability of life satisfaction compared with the reference region (Western). The associated differences between region of residence and life satisfaction among males and females are as follows. Compared to women residents of the Western region, women residents of Brong Ahafo, Eastern, Northern, Upper East, and Upper West regions had an increased probability of reporting higher life satisfaction levels. However, the relationship between residing in these regions on life satisfaction was not significant for males. Similarly, compared to men residents of the Western region, men residents of the Central region had a reduced probability of reporting higher life satisfaction. However, this relationship was not significant for females.

Predicted probabilities of associated factors of life satisfaction

For better interpretation, Table 3 supplements the multivariable oprobit results by reporting the average marginal effect of the predictor variables [44]. Because the dependent variable takes all levels of life satisfaction into account, the estimations of the marginal effect yielded 11 sets of results. However, only three points on the life satisfaction ladder namely lowest, middle and highest are presented to save space. The predicted probabilities were interpreted by comparing the probabilities of a variable’s reference category with the probabilities of other categories. From Table 3, the predicted probability of reporting higher satisfaction levels for males aged 15–19, 20–24, 25–29, 30–34, 35–39 and 40–44 years is 0.071, 0.051, 0.048, 0.060, 0.061 and 0.068, respectively, which were lower than the 0.074 of males aged 45-49years, holding all other variables constant. Similarly, the probability of reporting the highest life satisfaction for women aged 15–19, 20–24, 25–29, 30–34, 35–39 and 40–44 years was 0.121, 0.102, 0.101, 0.121, 0.115 and 0.102, respectively, lower than the 0.131 of women aged 45-49years. Lastly, compared with those without formal education or only earned pre-primary education, we find that having higher education on average increases the probability of identifying as highly satisfied with life by 0.09 for males and 0.17 for females. This finding conversely mirrors life satisfaction at its lowest level, where highly educated males (0.011) and females (0.014) had reduced average probabilities of reporting lowest life satisfaction compared to males (0.022) and females (0.028) with no or pre-primary education, holding all other variables constant. Full details of the other variables are reported in Table 3. The predicted probabilities for age groups, education level, wealth quintile and marital status at the highest levels of life satisfaction are plotted in S1 Fig.

Discussion

This study examined determinants of life satisfaction among Ghanaians aged 15 to 49 years. In the full sample multivariable model, gender, age, education, marital status, parity, wealth index, and region of residence were significantly associated with life satisfaction. Same relationships existed in the gender-stratified samples. However, there were some slight variations across genders. This will be the focus of our discussion.

Within the age groups, those who were aged 20–24, 25–29, and 40–44 years had a reduced probability to be satisfied with life compared with those within the age group of 45–49 years. We found that the results were similar for both young male and female adults. However, only older females aged 40–44 reported lower levels of life satisfaction. This age pattern (which is depicted in S1 Fig) clearly mirrors the U-shape well-being curve indicating that happiness declines from late adolescence and rises in midlife [7,46]. Transitioning from adolescence to early adulthood is a vulnerable period in which young people take their first tentative steps toward independence. This phase (20–29 years) is often associated with significant life changes and responsibilities as an individual works toward his or her goals including emancipation, getting married, getting a higher education, and employment [47]. Actualizing these goals tends to put pressure and stress on young individuals and severely impairs their well-being and life satisfaction [47,48]. In responding to life stressors, some young adults may engage in risky health behaviours such as substance use [49]. Frequent usage of substances such as smoking cigarettes, marijuana, or drinking alcohol which is commonest among young adults than older adults [50], increases the likelihood of low life satisfaction [51]. For females between the ages of 40 and 44 fertility declines and the onset of menopausal syndrome begins [52]. The desire to have children, as well as initial responses to these menopausal symptoms such as depression, hot flashes, and insomnia, may explain their decreases in life satisfaction [53,54].

Being highly educated was related to perceived higher life satisfaction for both men and women. This finding agrees with Powdthavee et al. [55] who reported higher levels of life satisfaction among the highly educated. According to Maslow’s hierarchy of needs [56] after the satisfaction of basic needs such as food, water, shelter, and clothing, the next higher human needs among others include higher forms of education. It is believed that the gratification of this need comes with higher levels of life satisfaction [56,57]. For instance, education serves as a springboard for better career opportunities and reduces the risk of being unemployed [58]. Job opportunities and higher incomes serve as indirect conduits through which education increases life satisfaction [57,59]. On the other hand, life dissatisfaction, anger, frustration, and unhappiness have been associated with unemployment [60]. This finding is consistent with previous research indicating that higher education is significantly related to the degree to which both men and women are satisfied with their lives [61].

We also found that married men and women were more satisfied with life than those who were previously married. While this contradicts the findings of Addai et al. [19], who found that marriage is negatively associated with happiness and life satisfaction in the Ghanaian context, it supports the findings of Botha and Booysen’s [62] study, which reports that married people in South Africa have higher levels of life satisfaction than divorced people. Marriage brings about love, mutual support, protection, social control, economic benefits, and all these tend to increase life satisfaction [63]. Marriage in Ghana is a social event guided by customary law and held in high esteem [19,64]. This is not surprising because a study by Dery and Bawa revealed that women in Northern Ghana were more satisfied with life mostly because marriage is revered by women and is associated with some form of prestige [65]. Furthermore, in Ghana, men are expected to marry at a certain age (young-middle adulthood) to be considered responsible and “men” [65,66]. Once this is done, some level of prestige is established, and societal pressure reduces. The study’s findings also revealed that unmarried men equally reported higher levels of satisfaction. The single status also has advantages: it is associated with fewer responsibilities, greater individual growth, and independence, greater freedom, more energy towards career goals, more time for friendship, fewer marital issues such as domestic fighting, and greater peace [67,68] and these contribute to an increase in life satisfaction.

Females with one and more children reported reduced life satisfaction compared to their counterparts who had no children. This finding is consistent with the findings showing that mothers having and raising children may experience reduced life satisfaction [69,70]. This is because children demand lots of attention and time. Having children involves providing care and performing household duties such as shopping, washing, cleaning, and cooking. This increased workload is especially true for Ghanaian women because women are not only responsible for catering for their children but work outside their homes in a variety of formal and informal occupations [71]. The increased demands of juggling between child-rearing and careers may account for the negative impact on life satisfaction among females with one or more children.

Furthermore, our findings revealed that men with health insurance coverage had an increased probability of reporting higher life satisfaction. Although this relationship was not significant among women, the general consensus is that health insurance coverage improves access to healthcare services and utilization by lowering medical costs [72]. This does not only reduce financial burden but increases health seeking behaviours, affords medical screening and prompt medical support in the case of poor health [73,74]. Health insurance coverage may therefore ensure that men are in good mental and physical health, leading to higher levels of life satisfaction [75-77].

Our results further revealed that household wealth is a significant determinant of life satisfaction. According to the findings, males in the richest wealth quintile and females in the fourth and richest wealth quintiles had higher levels of life satisfaction compared to those in the poorest wealth quintile. We argue that individuals from wealthy households can satisfy their physiological needs and are less concerned about financial burdens which increases their chances of achieving dreams and meeting higher-order needs and thus increasing life satisfaction in the long run [78,79]. In addition, individuals from wealthy families are happier and tend to have better social relationships, health, infrastructure, and leisure opportunities [78,79]. However, people living in poverty struggle with acquiring basic survival needs, leading to lower motivations to succeed in life [80]. Moreover, individuals from poor economic backgrounds are dissatisfied with their circumstances and faced financial difficulties such as increased debt and the inability to pay bills [81,82]. Due to these stressors, they are more likely to have low self-esteem, anxiety and frustration, and subsequently low life satisfaction [83,84].

Finally, being resident in the Greater Accra and Volta regions was associated with a higher chance of life satisfaction compared to the Western region for both men and women. Additionally, for women, being resident in Brong Ahafo, Eastern, Northern, Upper East, and Upper West regions was associated with higher life satisfaction. Although reasons for this relationship may not be fully known to the authors, a combination of factors including vast arable lands, agrarian activities, high levels of connectedness, and religiosity may be contributing to the life satisfaction of dwellers in the Brong Ahafo, Eastern, Northern, Upper East, Upper West and Volta regions [21,85-88]. The conditions may be different for the Greater Accra Region, which is predominantly urban, highly industrialized, and houses the capital city. It is a region that offers dwellers greater social, employment and economic opportunities, leading to higher levels of income and perhaps high levels of life satisfaction [89-91]. Our results inversely show that residents in the Western region were less likely to report being highly satisfied with life and this could be explained by the discovery of gas and oil in the Western belt of the country [92,93]. According to Arthur and Amo-Fosu [92] the discovery of gas and oil in the region led to higher costs of living such as increases in commodities and accommodation which has affected local residents. This has led to increases in worry among residents and may account for lower levels of satisfaction.

Strengths and limitations

One strength of this paper is its ability to stratify established relationships along gender lines, generating more richer information about the determinants of life satisfaction between men and women in Ghana. Another strength is the study’s use of a nationally representative dataset which facilitates generalization and enhances reliability by lessening the effects of potential errors induced by self-reporting. Nevertheless, these findings ought to be interpreted with caution due to limitations. First of all, the use of cross-sectional study limits the ability to assess the trends and also establish causation between the various factors and life satisfaction. It is therefore recommended that the associated factors explored in this study should be studied more longitudinally. Additionally, future Ghanaian studies should attempt using other robust analysis such as multilevel modelling as well as testing interaction effects (e.g., age-gender interaction). Secondly, we may have misspecified our model after excluding health variables (e.g., “Difficulty hearing, even if using a hearing aid”) of individuals, however, it was necessary since data on these variables were collected only from participants aged 18-49years. MICS should endeavour to include data on health variables for 15–19 years individuals in future datasets. We would also like to mention that our findings only extend previous literature on the subject matter of life satisfaction in Ghana.

Conclusion

Our research presents findings suggesting that gender, age, parity, education, marital status, wealth index, and region of residence are determinants of life satisfaction or thriving among Ghanaians. It is also reported that this pattern of relationships slightly varied between men and women. These findings collectively provide useful information for policymakers, researchers, and practitioners. For instance, policies designed towards providing services for the Ghanaian population to improve life satisfaction should be distributed equitably and equivalently across gender, taking into consideration the intricate relationships between determinants and life satisfaction as established in this study. Evidence from this study also calls on the government of Ghana to begin tracking the life satisfaction of her citizens. In recent times, the inclusion of self-reported well-being and life satisfaction in governmental policies for tracking objective social and economic progress has been advocated [94,95]. Because of this proposal, many nations and international development organizations have taken the necessary steps to make life satisfaction central to developmental policies [96,97]. The United Nations, for instance, has included “Good health and well-being” on its list of 17 Sustainable Development Goals [98-100]. Therefore, our findings provide a step towards this realization in Ghana.

Predicted probabilities of life satisfaction at highest level.

Margins plot with confidence intervals. Blue lines represent men and red lines represent women. Note: PP = Pre-primary education; predicted probabilities (on the y-axis) were derived from full sample model interacting with gender with age groups, education level, wealth quintile and marital status (each on the x-axis).

(TIF)

Click here for additional data file.

Intercorrelation between study variables.

(PDF)

Click here for additional data file.

Bivariate oprobit model regressing life satisfaction on predictor variables in the full sample and gender stratified sample.

(PDF)

Click here for additional data file.

Stata dataset of study.

Abridged version of the Stata dataset analysed for the study.

(DTA)

Click here for additional data file.

Stata do-file of study.

Stata do-file containing the commands used to run the statistical analyses.

(DO)

Click here for additional data file.

1 Apr 2021

PONE-D-20-37979

Determinants of life satisfaction among reproductive age Ghanaians: A further analysis of the 2017/2018 Multiple Cluster Indicator Survey.

PLOS ONE

Dear Nutifafa Eugene Yaw Dey,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The paper has without doubt the potential to be published

The conceptual objections of Reviewer 2 are quite easy to address.

Central are the methodological objections of Reviewer 1 as well as Reviewer 2:

- inclusion of happiness as a predictor:

- dichotomization of the outcome

See the detailed remarks by both reviewers.

Please submit your revised manuscript by May 15, 2021. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see:  http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at  https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

I hope very much you will accept this invitation for a major revision. In case of a revision I will send the paoer again to both reviewers.

We look forward to receiving your revised manuscript.

Kind regards,

Gert G. Wagner, Professor

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. In the methods section, please provide details regarding how the household wealth index was categorised.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This manuscript investigates the associations between life satisfaction and various of its predictors in Ghanaians and highlights some differences between the genders. This study is overall well-conducted, and I’m a big fan of this type of more descriptive study in which associations are reported without squeezing them into any particular narrative.

I’d normally write a much longer review, but I believe that there is one central issue with the analysis that weakens all results (see below, inclusion of happiness as a predictor). Once this issue has been fixed, I’m more than happy to re-read this interesting work and provide more details. (I think this reduces the workload on both sides: I don’t have to comment on issues that will likely be fixed once analyses have been re-run; the authors don’t have to write lengthy replies to issues that no longer apply.).

Best regards,

Julia Rohrer

(I sign all my reviews)

Major issues:

- inclusion of happiness as a predictor: I believe that including happiness as a predictor will bias all your estimates. Why is that? First, think about what “the effect of X on life satisfaction, controlling for happiness” ought to mean – happiness is extremely closely related to life satisfaction. What does it mean to be more satisfied without being happier? More rigorously, I believe that a sensible model would be that both happiness and life satisfaction are measures of people’s underlying overall assessment of their lives. Controlling for happiness will remove a lot of valid variability. Thinking about it another way: how satisfied you are with your life may affect how happy you are right now. Happiness is thus an outcome of life satisfaction, including it in a model will introduce collider bias. I’ve written about collider bias elsewhere, maybe you’ll find it helpful: Rohrer, 2018, https://journals.sagepub.com/doi/full/10.1177/2515245917745629. I’d consider any of the following solutions satisfactory: omit happiness as a predictor; combine happiness and life satisfaction into a more reliable indicator of well-being (see also point below on dichotomization of the outcome); analyze both as outcomes separately (as robustness check, but also maybe to find out whether these measures react differently to various predictors).

- dichotomization of the outcome: I’m opposed to the dichotomization of the outcome; I believe that this steps discards information that could be valuable. I guess there could be some concern that respondents are not fully using the scale in a “reasonable” manner. That may as well be possible (I think in representative samples, many people are overwhelmed by too many response options), but dichotomizing isn’t really going to fix this issue—it just imposes the assumption that all 0-6 responses are the same, and that all 7-10 responses are the same, discarding valuable information (I do believe that somebody who says 0 is likely less satisfied than somebody saying 6, but your analysis would discard this information). If people don’t use a scale efficiently, they may as well “accidentally” respond 7 instead of 6, so you still have misclassification.

There is another issue with analyzing a dichotomous outcome: You know have to decide on which scale to assess interaction. Currently, you’re only looking at Odds Ratios. I believe that many economists wouldn’t be satisfied with that, as they are very much in favor of evaluating interactions on a probability scale. As it happens, I just had a paper accepted covering this topic, you may find it helpful: Rohrer & Arslan (2021), preprint: https://psyarxiv.com/7fm2j/ (relevant section is the first one on the scale dependence of interactions). If you keep the binary outcome, I’d like to see an evaluation of interaction on the scale of the probability of being thriving (Odds Ratio is rather unintuitive, its only merit are some nice statistical properties). If you instead do a regular linear regression, that won’t be an issue. (I’m aware that doing a simple linear regression with an ordinal output isn’t optimal either; but I think the proper solution would be an ordinal model rather than dichotomization. But I do think ordinal models are rather involved and often not quite interpretable, so I wouldn’t want you to run one of those either).

Minor issues:

- abstract, “in a full and gender-stratified model”: when reading this for the first time, I was really confused what “full” was referring to. A “full model”? A “fully-stratified” model? I assume you mean to say that you did analysis in two ways, one time for the full sample, one time for gender subsamples.

- line 74, “the age of Addai’s et al. dataset was within 2005-2008 while Calys-Tagoe’s et al. research focused on older adults 50years and above.”: I don’t understand what is being said here (is this a comparison of dates with ages?)

- p. 8, bivariate analysis: I don’t think it’s a good idea to use significance in a bivariate analysis as a criterion for inclusion of a variable. I even think I have seen people writing about this (it’s prone to overfitting, and in any case a bivariate association doesn’t tell you whether a variable has a causal effect or not). I still think the bivariate analysis is nice for full transparency, so I’d simply delete the sentence using it as a rationale for inclusion in the full model.

- I do like that you report gender-stratified analyses

Reviewer #2: Review of

Determinants of life satisfaction among reproductive age Ghanaians: A further analysis

of the 2017/2018 Multiple Cluster Indicator Survey

The primary aim of this study was to examine life satisfaction in a sample of Ghanaians. The authors applied logistic regression models to cross-sectional data from the Ghana Multiple Indicator Cluster Survey Six to examine these associations of life satisfaction with a number of covariates. Results are taken to indicate that less than 40 percent of participants reported being satisfied with their life and that covariates such as age, gender, and education were associated with life satisfaction. The authors pursue an interesting research question using a most likely understudied population. However, I see major conceptual and also major empirical concerns.

CONCEPTUAL

1. Most importantly, the theoretical focus of the paper needs to be clarified (p. 3-5). To illustrate, the authors state the importance of studying life satisfaction in Ghana and list a number of potential underlying factors such as poverty, but don’t elaborate on potential underlying mechanisms and how all of these constructs are linked to one another. It would be very helpful to also conceptualize these theoretical assumptions in the study or reframe the introduction so that it becomes clear how life satisfaction and the covariates are linked to one another, what the role of the specific regional context is and how this can be embedded into the existing literature.

2. The literature review is incomplete. To illustrate, a number of studies have examined life satisfaction in adults (e.g., Blanchflower & Oswald, 2008; Diener, Suh, Lucas, & Smith, 1999; Pavot & Diener, 1993). Many of these studies have also examined the role of individual difference characteristics for life satisfaction (e.g., living status). Some of those studies in a longitudinal manner. It is thus necessary to further highlight what the present study adds to the existing literature. Please clarify.

METHODOLOGICAL

p. 5-6: I stumbled across the sample selection and study description. To illustrate, it should be clearer what the recruitment strategy for the sample was. For example, are people of the same household part of the sample? If so, data wouldn’t be independent from one another and the model would need to account for potential statistical interdependence. Alternative one would need to ensure to use only one member of the household.

p. 8: The authors should provide an intercorrelation table of all variables under study. This is important because it could provide further insight into the association of their related but ideally distinct outcome measure.

p. 8: Relatedly, I was surprised to see that the authors controlled for happiness when examining life satisfaction. From a conceptual perspective, both constructs are closely related to one another and might even be used interchangeably, depending on the measure that is used. I would also expect the constructs to be highly corrected. If not, this would hint at the measures examining something different but if so, this would result in multicollinearity issues.

p. 8: My biggest concern is that the authors dichotomized the outcome variable (struggling versus thriving) and applied logistic multiple regression analyses. In general, one should always make use of all data available. Artificially dichotomizing variables will result in a loss of information that would have been available otherwise. The authors have to make use of all data available by running at least a multiple regression analysis or ideally examine all variables multivariate framework (e.g., SEM). Please clarify.

p. 8: Relatedly, did the author test for any interaction effects? To illustrate, it could be that life satisfaction varies for older men but not older women (i.e., age-gender interaction) or for lower educated individuals living alone but high educated individuals living alone. Please clarify.

p. 9: Reporting percentages is not very informative. When examining a multivariate regression using continuous outcomes and predictors, the authors would need to report the regression coefficient and also effect sizes.

p. 9: Data on the geographical regions is part of the dataset. This is very interesting and could be of potential use for the authors. However, study participants are then nested in geographical regions and thus multilevel modeling should be applied.

MINOR ISSUES

p. 1: The authors switch between the terms well-being and life satisfaction throughout the manuscript. I would suggest sticking to one and provide a definition upfront.

p. 8: Were variables of interest standardized or centered (e.g., mean-centered). Please clarify.

e.g., p. 17: Throughout the manuscript, the authors imply directionality and sometimes causality in the associations between life satisfaction and the chosen covariates. I would encourage the authors to avoid any causal or directional language since the analysis does not allow to make any conclusions about causality or directionality.

p. 4: relatedly, since the data is cross-sectional the study does not examine ‘factors which contribute to the maintenance of a high level of life satisfaction in Ghana’ but ‘factors which are associated with life satisfaction in Ghana’.

Title, Abstract, p. 1-21: I wonder if it is necessary to add the term reproductive to the sample description when not discussing it further. E.g., would the authors expect this result to be an effect of reproductivity? Also, can one compare the reproductive age of men and women?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Julia M. Rohrer

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

27 May 2021

Response to Reviewers’ Comments

Manuscript Number: PONE-D-20-37979

Dear Editor,

Thank you for your letter and the opportunity to revise our paper on “Determinants of life satisfaction among Ghanaians aged 15 to 49 years: Further analysis of the 2017/2018 Multiple Cluster Indicator Survey”. We are grateful for the reviews provided by the reviewers.

Thank you for your continued interest in our research and we hope that this improved manuscript is accepted for publication in PLOS ONE. Below are our responses to the comments and concerns.

Editor comments

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Response

Thank you for the reminder. We have followed the formatting requirements accordingly.

2. In the methods section, please provide details regarding how the household wealth index was categorized

Response

The authors have provided these details. “Household wealth was categorized into poorest (0), second quintile (1), middle (2), fourth quintile (3), and richest (4).”

Reviewer comments:

Reviewer #1:

1. inclusion of happiness as a predictor: I believe that including happiness as a predictor will bias all your estimates. Why is that? First, think about what “the effect of X on life satisfaction, controlling for happiness” ought to mean – happiness is extremely closely related to life satisfaction. What does it mean to be more satisfied without being happier? More rigorously, I believe that a sensible model would be that both happiness and life satisfaction are measures of people’s underlying overall assessment of their lives. Controlling for happiness will remove a lot of valid variability. Thinking about it another way: how satisfied you are with your life may affect how happy you are right now. Happiness is thus an outcome of life satisfaction, including it in a model will introduce collider bias. I’ve written about collider bias elsewhere, maybe you’ll find it helpful: Rohrer, 2018, https://journals.sagepub.com/doi/full/10.1177/2515245917745629. I’d consider any of the following solutions satisfactory: omit happiness as a predictor; combine happiness and life satisfaction into a more reliable indicator of well-being (see also point below on dichotomization of the outcome); analyze both as outcomes separately (as robustness check, but also maybe to find out whether these measures react differently to various predictors).

Response

Thank you for noting these errors. As suggested, we have omitted happiness as a predictor.

2. Dichotomization of the outcome: I’m opposed to the dichotomization of the outcome; I believe that this steps discards information that could be valuable. I guess there could be some concern that respondents are not fully using the scale in a “reasonable” manner. That may as well be possible (I think in representative samples, many people are overwhelmed by too many response options), but dichotomizing isn’t really going to fix this issue—it just imposes the assumption that all 0-6 responses are the same, and that all 7-10 responses are the same, discarding valuable information (I do believe that somebody who says 0 is likely less satisfied than somebody saying 6, but your analysis would discard this information). If people don’t use a scale efficiently, they may as well “accidentally” respond 7 instead of 6, so you still have misclassification.

There is another issue with analyzing a dichotomous outcome: You know have to decide on which scale to assess interaction. Currently, you’re only looking at Odds Ratios. I believe that many economists wouldn’t be satisfied with that, as they are very much in favor of evaluating interactions on a probability scale. As it happens, I just had a paper accepted covering this topic, you may find it helpful: Rohrer & Arslan (2021), preprint: https://psyarxiv.com/7fm2j/ (relevant section is the first one on the scale dependence of interactions). If you keep the binary outcome, I’d like to see an evaluation of interaction on the scale of the probability of being thriving (Odds Ratio is rather unintuitive, its only merit are some nice statistical properties). If you instead do a regular linear regression, that won’t be an issue. (I’m aware that doing a simple linear regression with an ordinal output isn’t optimal either; but I think the proper solution would be an ordinal model rather than dichotomization. But I do think ordinal models are rather involved and often not quite interpretable, so I wouldn’t want you to run one of those either).

Response

Thank you for this suggestion. We have removed dichotomization of the outcome variable. Currently, we are using the variable in its original state, using all the points on the Cantril’s Self-Anchoring Ladder of Life Satisfaction scale. Correspondingly, the data analysis was changed from binary logistic regression to ordered probit regression as you suggested.

3. abstract, “in a full and gender-stratified model”: when reading this for the first time, I was really confused what “full” was referring to. A “full model”? A “fully-stratified” model? I assume you mean to say that you did analysis in two ways, one time for the full sample, one time for gender subsamples

Response

This correction has been done. Kindly find this correction on the abstract page. “This study, therefore, extends previous literature by examining the determinants of life satisfaction among Ghanaians in two ways: a full sample and gender-stratified sample.”

4. line 74, “the age of Addai’s et al. dataset was within 2005-2008 while Calys-Tagoe’s et al. research focused on older adults 50years and above.”: I don’t understand what is being said here (is this a comparison of dates with ages?)

Response

This correction has been done. Kindly find this correction on page 4. “The age of Addai et al’s (41) dataset is now quite old, using data that was collected within 2005-2008. Also, Calys-Tagoe et al’ study (12) only focused on older adults (50 years and above).”

5. p. 8, bivariate analysis: I don’t think it’s a good idea to use significance in a bivariate analysis as a criterion for inclusion of a variable. I even think I have seen people writing about this (it’s prone to overfitting, and in any case a bivariate association doesn’t tell you whether a variable has a causal effect or not). I still think the bivariate analysis is nice for full transparency, so I’d simply delete the sentence using it as a rationale for inclusion in the full model.

Response

Thank you for this suggestion. We have removed this sentence from the manuscript.

6. - I do like that you report gender-stratified analyses

Response

Thank you for the feedback.

Reviewer #2:

1. Most importantly, the theoretical focus of the paper needs to be clarified (p. 3-5). To illustrate, the authors state the importance of studying life satisfaction in Ghana and list a number of potential underlying factors such as poverty, but don’t elaborate on potential underlying mechanisms and how all of these constructs are linked to one another. It would be very helpful to also conceptualize these theoretical assumptions in the study or reframe the introduction so that it becomes clear how life satisfaction and the covariates are linked to one another, what the role of the specific regional context is and how this can be embedded into the existing literature

Response

Thank you for this suggestion. We have accordingly reframed the introduction by clearly indicate how life satisfaction and the covariates are linked to one another. Here is an example of text elaborating this on page 3. “In Diego-Rosell, Tortora, and Bird’s (35) study, for instance, it was found out that families with a high household income had better life satisfaction across the 153 countries. Luhmann et al. (36) also found from analysing longitudinal data from three nationally representative panel studies that life satisfaction levels increased with marriage and childbirth but reduced with marital separation, job loss, starting a new job, and relocating”.

2. The literature review is incomplete. To illustrate, a number of studies have examined life satisfaction in adults (e.g., Blanchflower & Oswald, 2008; Diener, Suh, Lucas, & Smith, 1999; Pavot & Diener, 1993). Many of these studies have also examined the role of individual difference characteristics for life satisfaction (e.g., living status). Some of those studies in a longitudinal manner. It is thus necessary to further highlight what the present study adds to the existing literature. Please clarify.

Response

Thank you for bringing this to our attention. We have further highlighted on how this present study adds to existing literature. See page 5, “Our study goes a step further by examining closely these factors from a gendered perspective; an examination that is virtually nonexistent in Ghana. Due to the variations in social norms and biological characteristics, a gendered perspective has been recommended to provide more nuanced information into the associated factors of both men and women’s life satisfaction (36–38).”

3. p. 5-6: I stumbled across the sample selection and study description. To illustrate, it should be clearer what the recruitment strategy for the sample was. For example, are people of the same household part of the sample? If so, data wouldn’t be independent from one another and the model would need to account for potential statistical interdependence. Alternative one would need to ensure to use only one member of the household.

Response

Thank you for this suggestion. We have clearly indicated the recruitment strategy for the sample. A complex survey design was used to account for statistical interdependence. See page 9, “In each household, people from the same household were selected.” and, page 6, “We accounted for the complex sampling design embedded in the dataset to monitor possible analytical errors and make proper inferences (45). This was achieved by correcting for clusters, stratification, and sample weights using the complex survey mode command 'svyset'.”

4. p. 8: The authors should provide an intercorrelation table of all variables under study. This is important because it could provide further insight into the association of their related but ideally distinct outcome measure.

Response

Thank you for bringing this to our attention. An intercorrelation table has been included as supplementary material. For easy identification of the intercorrelation table, a statement has been provided. See page 12, “Spearman’s rho was used as a preliminary test to assess the intercorrelations between all study variables (see S1 Table).”

5. p. 8: Relatedly, I was surprised to see that the authors controlled for happiness when examining life satisfaction. From a conceptual perspective, both constructs are closely related to one another and might even be used interchangeably, depending on the measure that is used. I would also expect the constructs to be highly corrected. If not, this would hint at the measures examining something different but if so, this would result in multicollinearity issues

Response

Thank you for noting this. As suggested, we have omitted happiness as a predictor.

6. p. 8: My biggest concern is that the authors dichotomized the outcome variable (struggling versus thriving) and applied logistic multiple regression analyses. In general, one should always make use of all data available. Artificially dichotomizing variables will result in a loss of information that would have been available otherwise. The authors have to make use of all data available by running at least a multiple regression analysis or ideally examine all variables multivariate framework (e.g., SEM). Please clarify.

Response

Thank you for this suggestion. We have removed dichotomization of the outcome variable. Currently, we are using the variable in its original state, using all the points on the Cantril’s Self-Anchoring Ladder of Life Satisfaction scale. Correspondingly, the data analysis was changed from binary logistic regression to ordered probit regression as you suggested.

7. p. 8: Relatedly, did the author test for any interaction effects? To illustrate, it could be that life satisfaction varies for older men but not older women (i.e., age-gender interaction) or for lower educated individuals living alone but high educated individuals living alone. Please clarify.

Response

There was no interaction effect tested in the new analysis. Although, the authors acknowledge the impact of such interaction, testing interaction effect was beyond the scope of the study’s objectives. Your suggestion has been recommended for future researchers. See page 21, “Additionally, future Ghanaian studies should attempt using other robust analysis such as multilevel modelling as well as testing interaction effects (e.g., age-gender interaction).”

8. p. 9: Reporting percentages is not very informative. When examining a multivariate regression using continuous outcomes and predictors, the authors would need to report the regression coefficient and also effect sizes.

Response

We have reported the regression coefficient and average marginal effect. These can be found in Table 2 and 3.

9. p. 9: Data on the geographical regions is part of the dataset. This is very interesting and could be of potential use for the authors. However, study participants are then nested in geographical regions and thus multilevel modeling should be applied.

Response

Thank you for this suggestion. Although we agree that multilevel modeling is one of the alternative methods, complex survey analysis also accounts for these inherent variations and clusters. We, therefore, have recommended the use of multilevel modelling for future studies in Ghana. See page 21, “Additionally, future Ghanaian studies should attempt using other robust analysis such as multilevel modelling as well as testing interaction effects (e.g., age-gender interaction).”

10. p. 1: The authors switch between the terms well-being and life satisfaction throughout the manuscript. I would suggest sticking to one and provide a definition upfront.

Response

We have accordingly sticked to the use of only life satisfaction in this manuscript. You can find these throughout the manuscript.

11. p. 8: Were variables of interest standardized or centered (e.g., mean-centered). Please clarify.

Response

The analysis was redone using ordered probit regression. Because ordered probit regression accounts for all the points in the outcome variable, there was no need for us to standardize or center the variable.

12. e.g., p. 17: Throughout the manuscript, the authors imply directionality and sometimes causality in the associations between life satisfaction and the chosen covariates. I would encourage the authors to avoid any causal or directional language since the analysis does not allow to make any conclusions about causality or directionality.

Response

Thank you for noting this. We have accordingly removed any causal language. Here is an example of text highlighting this change, on page 16, “The association between marital status, parity, household wealth index, region of residence, and life satisfaction were different for females and males.”

13. p. 4: relatedly, since the data is cross-sectional the study does not examine ‘factors which contribute to the maintenance of a high level of life satisfaction in Ghana’ but ‘factors which are associated with life satisfaction in Ghana’.

Response

Thank you for suggesting this. We have reframed our introduction and thus, this statement has been omitted from the work.

14. Title, Abstract, p. 1-21: I wonder if it is necessary to add the term reproductive to the sample description when not discussing it further. E.g., would the authors expect this result to be an effect of reproductivity? Also, can one compare the reproductive age of men and women?

Response

This correction has been made. Kindly find this correction on the abstract page “This study, therefore, extends previous literature by examining the determinants of life satisfaction among Ghanaians in two ways: a full model and gender-stratified model” and the title page, “Determinants of life satisfaction among Ghanaians aged 15 to 49 years: A further analysis of the 2017/2018 Multiple Cluster Indicator Survey.”

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

28 Jul 2021

PONE-D-20-37979R1

Determinants of life satisfaction among Ghanaians aged 15 to 49 years: A further analysis of the 2017/2018 Multiple Cluster Indicator Survey

PLOS ONE

Dear Dr. Dey,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the revision process.It is felt that the manuscript improved enormously from the previous drafts. Now the analysis is sound with the removal of happiness and the ordered probit.Reviewer 1 has minor suggestions to edit. There is sometimes a choice of innacurate words like rudimentary for elemmentary.The paper is missing a methods section describing the ordered probit model used explaining its strenghts and limitations in this context. Please submit your revised manuscript by Sep 11 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see:  http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at  https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

We look forward to receiving your revised manuscript.

Kind regards,

José Antonio Ortega, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: I’d like the authors for carefully addressing all of my comments. My previous concerns no longer apply to the revised version of the manuscript, which I believe will make a great contribution to the literature on well-being. I have only very few minor remarks left, such as typos or small tweaks to the language. Apart from that, I’d like to encourage the others to make their Stata do-files available for the scientific community, for example, on the Open Science Framework (osf.io). That way, other researchers can apply for access to the data used and then use the do-files to reproduce results; but it is also really helpful to double-check how exactly models were specified.

(I believe that this would also be aligned with PLOS ONE’s Data Availability Policy)

Best regards,

Julia Rohrer (I sign all my reviews)

p. 3, “life satisfaction across the 153 countries”: I think this should either be “across 153 countries” or “across the 153 countries included in their study.”

p. 4, “Considering these limitations and the pressing need to produce more recent evidence regardless of the persisting contextual problems ranging from limited access to drinking water [26], unemployment particularly among the youth [27], limited access to health care, poverty [28], high prevalence of chronic diseases [29-33] and poor quality education [34,35] that may threaten one’s life satisfaction, our study used the 2017/2018 Multiple Cluster Indicator Survey to examine the factors associated with life satisfaction in Ghana.”: I find the usage of “regardless of” here confusing. I wouldn’t say we need more evidence regardless of the problems, I would say we need more evidence in particular because of these problems?

p. 6, “The final samples were 660 clusters and 13202 households to the sampling strata.”: I’m confused by “to the sampling strata”; do you mean “across all sampling strata”?

p. 7, “This recategorized variable was kept for only descriptive purposes; therefore, the original variable (i.e., the 0 to 10 ordinal variable) was used in the main study analyses.”: I think this sentence actually becomes clearer if you just omit the “therefore”

p. 15, Table 3, 2 column from left, Parity Effects for Men at middle life satisfaction: there’s five cells with the same value (.195), I suspect this may be a copy-paste-error

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Julia M. Rohrer

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

8 Aug 2021

Dear Dr. José Antonio Ortega,

Thank you very much for giving us another opportunity to improve our manuscript. Below are the responses to the editorial and reviewer’s comments.

Editor’s comment

There is sometimes a choice of innacurate words like rudimentary for elemmentary.

Authors’ response

We have rectified this error. The new sentence reads like this: “Briefly, most of the selected variables were measured in a simple manner…”

Editor’s comment

The paper is missing a methods section describing the ordered probit model used explaining its strengths and limitations in this context.

Authors’ response

Thank you for noting this omission. We have updated the manuscript accordingly. Find this on page 8: “The ordered probit model is typically used to examine the variation in the data points of an ordinal categorical dependent variable (44). Though argued to produce parameter estimates difficult to interpret, oprobit was fitted mainly for its ability to preserve the ordering of the response options in the outcome variable as a function of the predictor variables (45).”

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Authors’ response.

The reference list has been double checked and we found no retracted papers. However, three more references were included in the reference list.

Reviewers' comment

Reviewer #1: I’d like the authors for carefully addressing all of my comments. My previous concerns no longer apply to the revised version of the manuscript, which I believe will make a great contribution to the literature on well-being. I have only very few minor remarks left, such as typos or small tweaks to the language. Apart from that, I’d like to encourage the others to make their Stata do-files available for the scientific community, for example, on the Open Science Framework (osf.io). That way, other researchers can apply for access to the data used and then use the do-files to reproduce results; but it is also really helpful to double-check how exactly models were specified.

(I believe that this would also be aligned with PLOS ONE’s Data Availability Policy)

Authors’ response

Thank you for the feedback and yes, the Stata do-file as well as the dataset are attached to the revised manuscript.

Reviewer’s comment

p. 3, “life satisfaction across the 153 countries”: I think this should either be “across 153 countries” or “across the 153 countries included in their study.”

Authors’ response

Thank you for pointing this out. We have accordingly corrected this sentence: “…it was found that families with a high household income had better life satisfaction across 153 countries.”

Reviewer’s comment

p. 4, “Considering these limitations and the pressing need to produce more recent evidence regardless of the persisting contextual problems ranging from limited access to drinking water [26], unemployment particularly among the youth [27], limited access to health care, poverty [28], high prevalence of chronic diseases [29-33] and poor quality education [34,35] that may threaten one’s life satisfaction, our study used the 2017/2018 Multiple Cluster Indicator Survey to examine the factors associated with life satisfaction in Ghana.”: I find the usage of “regardless of” here confusing. I wouldn’t say we need more evidence regardless of the problems, I would say we need more evidence in particular because of these problems?

Authors response

Thank you for noting this error and the suggestion. The sentence has been restructured as such: “Considering these limitations and the pressing need to produce more recent evidence in particular because of persisting contextual problems ranging…”

Reviewer’s comment

p. 6, “The final samples were 660 clusters and 13202 households to the sampling strata.”: I’m confused by “to the sampling strata”; do you mean “across all sampling strata”?

Authors’ response

Indeed, we meant “across all sampling strata” and this correction has been made. Thank you.

Reviewer’s comment

p. 7, “This recategorized variable was kept for only descriptive purposes; therefore, the original variable (i.e., the 0 to 10 ordinal variable) was used in the main study analyses.”: I think this sentence actually becomes clearer if you just omit the “therefore”

Authors’ response

This error has been corrected thanks to your suggestion. Here is the updated sentence: “This recategorized variable was kept solely for the purposes of description; the original variable…”

Reviewer’s comment

p. 15, Table 3, 2 column from left, Parity Effects for Men at middle life satisfaction: there’s five cells with the same value (.195), I suspect this may be a copy-paste-error

Authors’ response

Interestingly, this is not a copy-paste error. Similar values were generated through the margins command and rounded to .195. We are assuming they are the same because the probabilities are being predicted at the middle level of life satisfaction.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

29 Nov 2021

Determinants of life satisfaction among Ghanaians aged 15 to 49 years: A further analysis of the 2017/2018 Multiple Cluster Indicator Survey

PONE-D-20-37979R2

Dear Dr. Dey,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

José Antonio Ortega, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

It is deemed that the issues have been satisfactorily addressed and the article is ready for publication. Congratulations.

While it is not required, if you want you can add some paragraph on the difference of the age pattern found with other studies. The minimum satisfaction is generally at 45-55 years. Ghana seems different. (Eg: https://ideas.repec.org/p/pra/mprapa/7302.html but see also https://doi.org/10.1007/s00148-020-00797-z with cross-cultural evidence).

Reviewers' comments:

21 Dec 2021

PONE-D-20-37979R2

Determinants of life satisfaction among Ghanaians aged 15 to 49 years: A further analysis of the 2017/2018 Multiple Cluster Indicator Survey.

Dear Dr. Dey:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. José Antonio Ortega

Academic Editor

PLOS ONE